Detect every frame
Hey,
i have combined the rgb, depth frame with YOLO for an object detection task. My cpu can handle about 10 - 13 FPS. The video i recorded and stored as an .bag file has 25 FPS.
Problem is that when im open the video with the pyrealsense2 package for python only every second frame gets execute (because of the slow CPU). Can i slow down the video pipeline so every image is processed? Or is their another way to execute every picture?
It wouldn't be a problem if the video is then only half of the speed.
Thank you very much :)
-
Hi Andi Forster My first thought for an approach to this problem would be to alter the frame queue size to add latency and reduce the risk that frames are dropped. Changing the frame queue size affects the total number of frames in circulation.
https://dev.intelrealsense.com/docs/frame-management#section-frame-drops-vs-latency
The default frame queue size is '1'.
A discussion at the link below may be a helpful reference for making a frame queue size change in Python.
Please sign in to leave a comment.
Comments
1 comment